Abstract: Outdoor positioning mainly relies on the GPS system, but GPS signals can be severely affected indoors and in dense building group. Many indoor areas have coved with WiFi due to the development of WLAN technology, so no additional beacons are needed for positioning. However, due to the complex indoor environment and the susceptibility of WiFi signals, indoor positioning based on WiFi fingerprints is facing great challenges. Therefore, we proposed a method to identify the buildings and its floors based on SCAE-DNN. The Stacked Contractive Auto-Encoder network (SCAE) is added to the Deep Neural Network (DNN). In the offline phase, SCAE is used to extract the characteristics of the WiFi signal strength as the input to the DNN. In the online phase, the network is trained to identify buildings and floors. The SCAE-DNN network is robust to the fluctuations of input signal, so the recognition accuracy is high. By using the public data set UJIIndoorLoc, the results shows that the accuracy achieves 99.7% by utilizing SCAE-DNN.
Keywords: Indoor location, WiFi location fingerprint, Deep Neural Network, Stacked Contractive Auto-encoders.
Title: Identification of buildings and floors based on SCAE-DNN
Author: Chunkai Chen, Baocong Liu
International Journal of Computer Science and Information Technology Research
ISSN 2348-1196 (print), ISSN 2348-120X (online)
Research Publish Journals
